necessary steps to do before starting a rollforwarddb

[repostor]

Hi,

I am struggling for; which steps are needed to do before starting the "rollforwarddb" command.

Eg;

Backup:
------------

1. Full Backup of database: ckpdb <database>
2. Backup files located here: dmp/default/<database>
3. Backup journal files located here: jnl/default/<database>

Restore:
------------

1. Create database: createdb <database>
2. Restore files located here: dmp/default/<database>
3. Restore journal files located here: jnl/default/<database>
4. Remove all files under data/default/*
5. Start a recover database, rollforwarddb <database>

Or is there any better instruction how to do a restore, and all the important files needed for a complete restore.

If I start the restoration without doing a "createdb" command, the restore failes.?

Please help me...

Regards Tomas

[nick.bolton]

The rollforwarddb command was designed to recover backups on the machine they were created on - hence it assumes that the database environment already exists.

Recovering backups on a DR environment requires you to fool Ingres into believing this is the machine that the backup came from - hence you also need to have the same sort of environment setup on the DR server (e.g. have issued createdb).

Ingres now offers incremental rollforwards, which has little obvious benefit outside of a DR scenario; but they leave it up to you to work out how to get the files over to DR.

I would recommend watching this presentation about incremental rollforward to get an overview of the checkpoint migration technique:
http://downloads.ingres.com/online/m...ollforward.swf

[chipn]

Hello Tomas,

Before you try a rollforward you want to make sure that the checkpoint was valid. You do that by running the "infodb <database-name>" command, searching for the checkpoint section (usually about 1/4 to 1/3 down from the top), and seeing that the status for the last checkpoint was "1" (a "0" means the checkpoint failed).

Next, you really want to make sure Ingres is shutdown and get a valid OS backup of all data locations. The reason for that is because one of the first things rollforwarddb does is remove files in the data directories. If rollforwarddb fails for some reason (such as an invalid checkpoint) then you can be in a lot of trouble if you haven't done this OS backup. The restore occurs in the same database, so you don't need to create a new database.

It is good to copy the data for iitables, iifile_info, and ii_multi_locations (just using the "select * from <table>" command in "sql") because this can help you manually remap tables and indexes to their underlying files if ever needed. I recommend getting this information at least weekly and saving it with the backup saveset.

Here is a good backup & recovery overview presentation given by Peter Gale (now a Principal Consultant with UK Operational Services at Ingres) a few years ago. It's a great "makes you think" kind of presentation. http://www.comp-soln.com/backup_presentation.pdf

Cheers,

Chip Nickolett

[repostor]

Thanks,
I will take a look at thoose links.